Well, it does. In another paper published in the Evostar conference, we compare several methods for measuring how good a combination is when compared to the others that could possibly be the solution; so far we had mostly used most parts (counting the number of non-zero partitions), but, in this paper, that compares our previous Evo method with another created by the coauthors, Maestro-Montojo and Salcedo-Sanz, we find that Entropy, at least for these sizes, is the way to go. Here’s the poster
You can access the paper Comparing Evolutionary Algorithms to Solve the Game of MasterMind, by Javier Maestro-Montojo, Juan Julián Merelo and Sancho Salcedo-Sanz (first and last authors from the University of Alcalá de Henares) online or request a copy from the authors.
Do you apply bioinspired techniques to games? Specially evolutionary algorithms? Submit a paper with your work to the Special Issue of Evolutionary Intelligence focused on games.
Anything related to games is valid, provided evolutionary techniques (or, in general, bioinspired techniques):
- Level design
- NPC design
- Learning in games
- Behaviour mining
- Puzzle solving
Please check the dates: the deadline for submission is May 10th, 2013, the notification of acceptance/revision/rejection will be given by June 10th, 2013 and the final manuscript submission date is July 1st, 2013. The articles must be submitted via the
As a result of a collaboration with Mario García Valdez, Leonardo Trujillo and Francisco Fernández (this one from Spain) we have published two papers based on the EvoSpace framework a pool-based evolutionary architecture for interactive and straight evolutionary computation. The first paper describes the EvoSpace-i, the interactive part and is well described by Paco Fernández in our group blog, and the
Hola que ases. Six projects have taken part this year in the 5th hackathon at Granada for the University Free Software Contest, three of which come from the GeNeura team. We spent all the weekend working at Cocorocó’s facilities. We worked but also enjoyed it and had fun. And the projects from GeNeura team are….:
OSGiLiath (from Pablo García, source code)
OSGiLiath (OSGi Laboratory for Implementation and Testing of metaHeuristics) is an open source framework for Service Oriented Evolutionary Algorithms.
ZomBlind (from Antonio Fernández, source code)
Zomblind is a post-apocalyptic mobile game of zombies. You have surely “lived” that, but now you can’t see your enemies, just hear them… Designed for players with visual impairments.
Code-Reimagined (from Javier Asensio, source code)
The idea is simple: turn your java code into a Super Maryo game stage. Then you can follow Maryo while debugging your code and quickly access any place on it. Many features are expected to be developed in the future.
The other competing projects are:
If you like any, join!
Clojure es un miembro de la familia LISP que se ejecuta sobre la JVM pudiéndo consumir y producir para dicho entorno.
Se destaca por poseer conceptos que facilitan el desarrollo de aplicaciones concurrentes además de la flexividad que de por sí significan las macros de los LISPs. En esta presentación se hace una introducción que destaca algunas de sus características así como ejemplica mediante demos.
Para profundizar en el tema se recomienda el libro Clojure Programming (2012) de Chas Emerick, Brian Carper, and Christophe Grand así como las presentaciones hechas por su creador:
Our journal paper on the Sandpile Mutation operator for Genetic Algorithms is now available online: C.M. Fernandes, J.L.J. Laredo, A.C. Rosa, J.J. Merelo, “The sandpile mutation Genetic Algorithm: an investigation on the working mechanisms of a diversity-oriented and self-organized mutation operator for non-stationary functions“, Applied Intelligence, February 2013.
Abstract: This paper reports the investigation on the sandpile mutation, an unconventional mutation control scheme for binary Genetic Algorithms (GA) inspired by the Self-Organized Criticality (SOC) theory. The operator, which is based on a SOC system known as sandpile, is able to generate mutation rates that, unlike those given by other methods of parameter control, oscillate between low values and very intense mutations events. The distribution of the mutation rates suggests that the algorithm can be an efficient and yet simple and context-independent approach for the optimization of non-stationary fitness functions. This paper studies the mutation scheme of the algorithm and proposes a new strategy that optimizes is performance. The results also demonstrate the advantages of using the fitness distribution of the population for controlling the mutation. An extensive experimental setup compares the sandpile mutation GA (GGASM) with two state-of-the-art evolutionary approaches to non-stationary optimization and with the Hypermutation GA, a classical approach to dynamic problems. The results demonstrate that GGASM is able to improve the other algorithms in several dynamic environments. Furthermore, the proposed method does not increase the parameter set of traditional GAs. A study of the distribution of the mutation rates shows that the distribution depends on the type of problem and dynamics, meaning that the algorithm is able to self-regulate the mutation. The effects of the operator on the diversity of the population during the run are also investigated. Finally, a study on the effects of the topology of the sandpile mutation on its performance demonstrates that an alternative topology has minor effects on the performance.